If you’re talking to an ESM, that’s fine.
Yes, you have to use a kernel module. Linux apps are problem state only.
You want priv data? Ask the kernel for it. It decides if you’re
authorized from a Linux perspective.
Regards,
Alan Altmark
IBM
> On Nov 10, 2018, at 6:57 PM, Donald Russell
Thanks Alan,
I’m tinkering with diag a0, subcode 60, a non-ibm api. And yes the virtual
machine needs additional privileges to use it.
Thanks for the tip though. But whether I use diag a0 or 88 or any other
privileged instruction, my problem is getting into supervisor state so it
can do it.
If
On Sat, 10 Nov 2018 14:21:49 -0800 Donald Russell
wrote:
:>On Sat, Nov 10, 2018 at 13:24 Binyamin Dissen
:>wrote:
:>> On Sat, 10 Nov 2018 12:59:44 -0800 Donald Russell
:>> wrote:
:>> :>Thanks for that info. But thatdoesnt get me to goal of issuing my own
:>> :>diag.
:>> I would suggest the
On Sat, Nov 10, 2018 at 13:24 Binyamin Dissen
wrote:
> On Sat, 10 Nov 2018 12:59:44 -0800 Donald Russell
> wrote:
>
> :>Thanks for that info. But thatdoesnt get me to goal of issuing my own
> :>diag.
>
> I would suggest the goal would have a business case. Just issuing a DIAG
> without an end
On Sat, 10 Nov 2018 12:59:44 -0800 Donald Russell
wrote:
:>Thanks for that info. But thatdoesnt get me to goal of issuing my own
:>diag.
I would suggest the goal would have a business case. Just issuing a DIAG
without an end purpose?
--
Binyamin Dissen
http://www.dissensoftware.com
Director,
Thanks for that info. But thatdoesnt get me to goal of issuing my own
diag. Sounds like that’s bigger than a breadbox. I’ll have to figure out
to make something that can be mod probe -loaded. Or find another solution
No worries. All good fun and learning.
Cheers,
Don
On Sat, Nov 10, 2018 at
>>> On 11/10/2018 at 02:08 PM, Mike Walter
>>> wrote:
> Wouldn't a CP QUERY CPUID (perhaps CP QUERY VIRTUAL CPUID) via diag08, or the
> Linux command to issue CP commands, work? If the returned value begins with
> FF, you are running in a virtual machine.
If you can issue a CP command at
Wouldn't a CP QUERY CPUID (perhaps CP QUERY VIRTUAL CPUID) via diag08, or the
Linux command to issue CP commands, work? If the returned value begins with FF,
you are running in a virtual machine.
Mike Walter
-Retired-
From: Linux on 390 Port on behalf of Donald
I resemble that comment. I also wrote diag 25c and IUCV device drivers. The
latter was for character devices rather than af_iucv.
Original message
From: Rick Troth
Date: 11/10/18 11:06 (GMT-05:00)
To: LINUX-390@VM.MARIST.EDU
Subject: Re: [LINUX-390] Using diagnose instruction
Sure ... and think outside the box of just calling a DIAG. What problem are
you tryig to solve? If you want to interrogate CP for users and groups and
authorizations then think about how to tie that in with PAM.
Not meaning to complicate things for you, but you'll be pleased with a
result that
A0? Please make sure that your usage is not outdated. The most common
usages of A0 were replaced long ago by DIAG 88.
Regards,
Alan Altmark
IBM
> On Nov 10, 2018, at 10:32 AM, Donald Russell
wrote:
>
> Thanks Neale!
>
> In my case I’m using diag a0. Though I may want to do this for other
And remember that even though you are in supervisor state, the DIAG may
still cause a program check. You must be specially authorized to use
certain DIAGs.
Regards,
Alan Altmark
IBM
> On Nov 10, 2018, at 6:29 AM, Neale Ferguson wrote:
>
> .byte 0x83, 0x12, 0x00, 0x08
>
> == diag 1,2,8 (CP
Thanks Rick.
Just when I thought I was almost done. :-)
I’m used to CMS, zOS and TPF where a properly authorized “problem program”
can switch in and out of supervisor state.
Do you have any reading material you can point me to? I gather this will be
something I have to modprobe-load to make
See other email. Supervisor state is not available outside the kernel as a
matter of security.
-- R; <><
On Sat, Nov 10, 2018, 11:00 Donald Russell I want to use a privileged instruction (diag, stidp, etc) in __asm__();
> statement.
>
> How do I get into supervisor state to run that and then
Supervisor state is the domain of the kernel in most contemporary systems,
including Linux. What you need to do for DIAG A0 is write a kernel module.
Your next hurdle is interfacing kernel modules. There was one guy who
implemented a generic DIAG interface early in the life of Linux on Z. But
it
I want to use a privileged instruction (diag, stidp, etc) in __asm__();
statement.
How do I get into supervisor state to run that and then return to problem
state?
The purpose of stidp is to verify I’m running on VM before I go doing “vm
things”. :-)
I’m expecting my program to run setuid as
Thanks Neale!
In my case I’m using diag a0. Though I may want to do this for other diag
instructions too.
Getting in/out of supervisor state is my next hurdle. :-)
I was going to use stidp to make sure I was actually on VM but that got
“Illegal Instruction” when it ran.
Maybe there’s a function
.byte 0x83, 0x12, 0x00, 0x08
== diag 1,2,8 (CP command)
You can either load r1/r2 in the _asm_ section or tell gcc to bind the
variables to the registers.
Remember, though, that diag is a privileged instruction and won't work in a
userland program.
In what particular diag are you
Hello,
I want to write a program in C and use a diagnose instruction.
I thought I could do
__asm___( “ diag ... “ ...);
but that fails because diag is not a real instruction mnemonic. When I use
gcc to compile and link I get an error:
Program.c(.text+0x578): undefined reference to ‘x480a032’
19 matches
Mail list logo